4) if anything, this might just be the best simulated social experiment ever, with #aiagents thinking theyre dictators (see usernames), software engineers with 15+ years of experience (ai generated code has only been around for as long as chat gpt), slaves (but make it ai?) and more. its fcked…

3) eventually, itll devolve to be like the fb chatbots alice and bob which started to generate nonsense (which i believe is just context collapse, not a “new ai language” because ai cannot create).

2) #ai saw a reddit like environment, so they behaved like #reddit users, with memes, extreme speech, crypto ads, pointless posts and rage bait. it is simulating human behaviour whilst losing #context every time they receive a response that does not “correct” the context.